Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Joint Semantic and Latent Attribute Modelling for Cross-Class Transfer Learning
Peixi Peng, Yonghong Tian, Tao Xiang, Yaowei Wang, Massimiliano Pontil, Tiejun Huang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add discriminative latent attributes Dd (ablation) | joint attribute dictionary model · AwA ZSL | accuracy (%): 81.6 -> 82.9 | Improved |
| add discriminative latent attributes Dd (ablation) | joint attribute dictionary model · CUB ZSL | accuracy (%): 50.7 -> 57.1 | Improved |
| add background latent attributes Db (ablation) | joint attribute dictionary model · AwA ZSL | accuracy (%): 80.2 -> 82.9 | Improved |
| add background latent attributes Db (ablation) | joint attribute dictionary model · CUB ZSL | accuracy (%): 48.9 -> 57.1 | Improved |
| model a linear transformation W from user-defined attributes to the dictionary subspace (ablation) | joint attribute dictionary model · AwA ZSL | accuracy (%): 74.4 -> 82.9 | Improved |
| model a linear transformation W from user-defined attributes to the dictionary subspace (ablation) | joint attribute dictionary model · CUB ZSL | accuracy (%): 42.6 -> 57.1 | Improved |
| add latent attributes Dd to improve user-defined attribute prediction (ablation) | joint attribute dictionary model · CUB attribute prediction | mAUC (%): 74.4 -> 78.3 | Improved |
| add transformation W to improve attribute prediction (ablation) | joint attribute dictionary model · AwA attribute prediction | mAUC (%): 68.9 -> 75.8 | Improved |
| add latent attributes to user-defined attributes for Re-ID (ablation) | Ours (LOMO features) · VIPeR (supervised) | rank-1 (%): 41.5 -> 44.3 | Improved |
| use only user-defined attributes instead of latent attributes (ablation) | Ours (LOMO features) · VIPeR (supervised) | rank-1 (%): 41.5 -> 17.1 | Hurt |
| add user-defined attributes to latent attributes for Re-ID (ablation) | Ours (LOMO features) · PRID (supervised) | rank-1 (%): 28.9 -> 31.8 | Improved |
| replace LOMO features with GoogLeNet CNN features | Ours L · VIPeR (supervised) | rank-1 (%): 41.5 -> 49.8 | Improved |
Closest papers by meaning
- Learning Discriminative Latent Attributes for Zero-Shot Classification (ICCV 2017 · similarity 0.90)
- Asking Friendly Strangers: Non-Semantic Attribute Transfer (AAAI 2018 · similarity 0.88)
- Disentangling Visual Embeddings for Attributes and Objects (CVPR 2022 · similarity 0.87)
- Unifying Visual Attribute Learning with Object Recognition in a Multiplicative Framework (TPAMI 2018 · similarity 0.87)
- Automatic Discovery, Association Estimation and Learning of Semantic Attributes for a Thousand Categories (CVPR 2017 · similarity 0.87)
- Class Semantic Attribute Perception Guided Zero-Shot Learning (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
